An initial list of symptoms potentially indicating ED in dogs and cats was drafted based on personal experience and previous literature (6, 12, 14, 15). Following consensus of six veterinary anesthesiologists (two residents and four diplomates of the European College of Veterinary Anesthesia and Analgesia, ECVAA), six symptoms were selected to be used in the clinical evaluation of ED occurrence: nystagmus, paddling, opisthotonos, uncoordinated movement, vocalization and biting. Additionally, it was established that a minimum of four symptoms had to be present to classify the phenomenon occurring during the recovery phase as ED.

Data were collected from clinical cases admitted at the small animal clinic of the University of Bern, between May 2022 and January 2023. An ad-hoc online form made available via a QR code was used (see Supplementary material). Staff members performing general anesthesia were asked to fill it out for each anesthetic event. For the present study, no specific ethical permission was deemed necessary by the responsible authority for animal experiments of the Canton of Bern, Switzerland.

Information was gathered for the following categories:

Data collected through the online forms were directly exported into a Microsoft Excel file.

For the descriptive analysis, dogs and cats were categorized into two groups, depending on the number of symptoms observed: animals showing 0–3 symptoms were assigned to the NED group (No-ED group) and animals showing 4–6 symptoms were assigned to the ED group.

The present study employed the penalized maximum likelihood logistic regression method proposed by Firth for rare events (16), utilizing a permissive p-value threshold of > 0.20 for univariable screening of various explanatory variables. Continuous variables were assessed for linear relationships with the log odds of the outcome using the Lowess smoothing function. In cases when linearity assumptions were not met, variables were stratified based on biological rationale and population distribution for further analysis. Predictors with at least 10 occurrences were retained for analysis. Backward elimination was performed, and the likelihood-ratio test was used to assess the contribution of any subset of parameters. The final multivariable model was constructed, with all variables having a statistical significance at p ≤ 0.05. If a variable affected another variable’s coefficient by more than 20%, it was considered a confounder and was forced into the model. All interactions were tested for significance. Significant variables were expressed as odds ratios with corresponding 95% Wald intervals and p values. The Hosmer–Lemeshow statistic was employed to evaluate the overall fit of the model. Given the rarity of the outcome and the limited sample, the feline and canine populations were then analyzed together, including only predictors meeting inclusion criteria in either species during univariate analysis. The need for restraint at recovery, sedative use prior to anesthesia, preventive sedation at recovery and ASA physical status classification was retested with univariable analysis. The same criteria previously discussed were used to build and test the multivariable model, ensuring the robustness and generalizability of the findings.

Detailed tables containing the demographic data can be found in the Supplementary material.

Among dogs, 136/139 (98%) had an ASA status of 3 or lower. Six were Border Collies: four received sedation during the recovery phase, 3 of which in a preventive way. Twenty dogs of the Retriever breeds (Golden Retriever, Labrador Retriever) were also included: six of them (30%) received sedation during the recovery phase, only one in a preventive way. Most of the dogs needing physical restraint during the recovery phase were allocated in the NED group 18/23 (78%); the majority of them (13/23; 57%) received a sedative drug.

Among cats, 44/45 (98%) had an ASA status of 3 or lower. Only 7/45 (16%) were sedated, 4 of which received it preventively. Seven cats out of 45 (16%) required physical restraint; most were allocated in the NED group (5/7; 71%).

The overall occurrence of ED symptoms is reported in Figure 1, while in Figure 2 symptoms occurring in the ED group only are reported. Figure 3 illustrates the percentage of animals showing 0–6 symptoms over the whole study population; 10/139 (7%) of the dogs and 7/45 (16%) of the cats (17/184 (9%) of the total population) showed all six symptoms.

Following univariable analysis, whether the dog received any sedation before presentation for anesthesia (Wald = 1.83, p = 0.18) and the need to restrain the animal (Wald = 3.03, p = 0.08) were the only variables meeting the criteria for inclusion in the multivariable model. However, the final model did not reach significance (Wald = 3.97, p = 0.14).

When considering the cat dataset, whether the cat’s ASA physical status was ≥ III (Wald = 1.65, p = 0.20) and the preventive use of postoperative sedation (Wald = 2.92, p = 0.09) met the criteria for inclusion in the multivariable model. However, they lost significance in the model (Wald = 3.82, p = 0.15). When considering both populations together, we found no association in the univariable analysis.

Within the study population, 55% (76/139) of the dogs and 62% (28/45) of the cats showed at least one symptom, but only 12% (16/139) dogs, and 22% (10/45) cats were categorized in group ED. Our results differ from those of Jones et al. (21), who reported a low recovery score in about one-third (29.1%) of the dogs. However, a direct comparison cannot be made due to the different approaches evaluating recovery quality. In particular, Jones et al. (21) used a simple descriptive scale (SDS) [as previously described by Lozano et al. (20)] and the level of Consciousness-Richmond Agitation and Sedation Scale (LOC-RASS) [as previously described by Sessler et al. (24)]. In the present study, a purpose-made scale was applied.

It is commonly suggested that certain dog breeds may experience rougher recovery phases. However, limited evidence supports this claim. According to a study by Kongara et al. (12), Alaskan Malamutes, Siberian Huskies and Labrador Retrievers are at higher risk of experiencing dysphoria due to the prevalence of a specific allele variant (C-15 A). However, Jones et al. (21) did not confirm this hypothesis, nor did the results presented here (only 20% of the Retriever dogs were in the ED group). Nonetheless, the limited number of animals per breed in this study poses challenges in interpreting the results. Further investigations with more animals should be conducted to identify possible breed predispositions.

In human medicine, there is no agreement on whether sex influences the prevalence of ED. While some studies report an increased prevalence in male individuals (14, 19, 22, 25, 26), others do not support it (27). In veterinary medicine, Jones et al. (21) could not find any evidence suggesting that sex influences recovery quality. Similar results were found in the present study.

Age has been recognized as a factor contributing to the development of ED in humans, with a higher likelihood present in pediatric patients (5, 22, 28, 29). The specific reasons for the increased incidence in children remain unclear. Some authors proposed that the combination of preschool age and the use of inhalants might act as a risk factor for ED. (1, 30, 31) Moreover, a higher occurrence of ED in preschool children compared to their older counterparts has also been suggested to be linked to the higher distress young individuals have in waking up in an unfamiliar environment (28, 30), which could resemble the situation occurring in small animals. In dogs, Jones et al. (21) could not find any evidence supporting the potential contribution of age to the development of poor recovery quality. In our study, animals in the ED group (both dogs and cats) were younger, but no statistical or clinically meaningful difference was found.

Different studies in human medicine investigated the possible correlation between ASA status and the appearance of ED; however, none could be confirmed (30–32).

Jones et al. found that an ASA status of III or higher was associated with a decrease in the incidence of poor quality of recovery (21). The authors suggest that this could be due to the slower metabolization of the anesthetics in these patients, allowing a slower and calmer recovery phase (21). In the present study, only four animals (3 dogs and 1 cat) with an ASA status above III were included, precluding further evaluation of this condition.

A factor shown to favor ED occurrence in children is anxiety (e.g., sleep disturbance, eating disturbances, aggression against authority) during the preoperative phase (33). In the present study, no clear link between preoperative behavior and the occurrence of ED symptoms was found.

Pain could also affect the recovery phase. There is no evidence in the literature, but distinguishing between painful conditions and ED during the recovery phase can be difficult. Methods that could help to differentiate pain from emergence delirium may include the use of pain assessment tools—such as the Visual Analogue Scale (VAS) employed in our study—or exclusion-based therapy through the administration of analgesic medications. However, distinguishing between the two phenomena remains challenging. In veterinary medicine, Jones et al. could not show any statistical significance between poor quality of recovery and painful procedures (21). Similarly, in our study, preoperative pain (as assessed by the VAS) did not seem to be linked with the development of ED (77% of the patients who showed ED had a VAS between 0 and 3).

In human medicine, benzodiazepines have been reported to increase the risk of ED (34, 35), and one study in veterinary medicine seems to support this hypothesis (10). The administration of opioids has also been reported to be linked to a worse recovery quality (12). However, the word dysphoria is often used in this case, referring to a phenomenon called “opioid dysphoria,” not further characterized in small animals (12, 13). In the present study, most animals received opioids, while only a few received benzodiazepines; thus, any attempt to statistically evaluate the contribution of these drugs to ED occurrence would be uninformative.

The use of TIVA has been reported to be a protective factor for the development of ED in humans when compared with sevoflurane (19, 34, 36). One explanation for this could lie in the pharmacokinetics of the injectable anesthetics (propofol), which could lead to a slower recovery phase and, thus, to a reduced occurrence of ED. (28) In our study, no association between TIVA and the occurrence of ED could be demonstrated. However, this is not surprising due to the low number of animals undergoing TIVA in this study.

Prolonged anesthesia duration seems to predispose to ED in humans (1, 34, 37). To the author’s knowledge, no such evidence is present for small animals, but a clear association between anesthesia duration and compromised recovery has been described in horses (38, 39). No association between the duration of anesthesia and ED occurrence has been found in our study.

Preventive sedation, administered before the recovery phase, aims to mitigate the risks associated with rapid recoveries, such as self-injury or injury to medical staff. Primarily, alpha-2 agonists such as dexmedetomidine or acepromazine were used for preventive sedation in the present study. The most common reasons for administering preventive sedation included procedures where the occurrence of emergence delirium during recovery could pose a risk to the animal (e.g., ophthalmologic procedures), or specific breed considerations (at the discretion of the anesthetist in charge). In human medicine, rapid recovery from anesthesia can increase the likelihood of postoperative complications, including ED in children (18, 21). A smoother transition from anesthesia to consciousness can be obtained by administering preventive sedation, which is hypothesized to lead to a better recovery quality (21). While guidelines for small animals are lacking, a recent study in dogs showed better recovery quality after a preventive administration of low-dose dexmedetomidine compared to saline (40). Preventive sedation could help to reduce ED, but further studies should be conducted before drawing conclusions.

In this study, 15/133 (12%) dogs and 3/44 (7%) cats were sedated preventively. Those animals were not included in further statistical analysis as we considered preventive sedation to impact recovery quality. Interestingly, the majority of the sedated animals did not show ED occurrence (14/15).

In veterinary medicine, the investigation of emergence delirium (ED) is still in its early stages. Its study is pivotal not only to improve the immediate post-operative period, but also to investigate on the potential long-term consequences. This is great concern also in human medicine (41), and could represent an interesting avenue for future research.